Body-aware robot, method and device for determining work sequence, and storage medium

Abstract

The application relates to the technical field of embodied intelligent industrial application, in particular to an embodied intelligent robot, a work sequence determination method and device and a storage medium. The method comprises the following steps: constructing a constraint parameter set according to steel bar binding work data; decomposing a global binding task into a plurality of independent intersection point work units according to steel bar deployment positions, and giving corresponding work weights according to the constraint parameter set; taking each intersection point work unit carrying the work weight as a task node, constructing a weighted semantic task graph among the task nodes according to the spatial position relationship among the intersection points; dividing the weighted semantic task graph into a plurality of subgraphs, and planning and solving the work sequence in each subgraph based on a graph search mode; and connecting the subgraph work sequences based on the boundary task nodes among the subgraphs to form a global work sequence. The method can solve the planning problem of the steel bar binding work sequence, and realize efficient steel bar binding work with industry standards as constraints.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of embodied intelligent industrial application, in particular to an embodied intelligent robot, a work sequence determination method and device, a computer readable storage medium and a computer program product. BACKGROUND

[0002] Embodied intelligence is an important branch of artificial intelligence, which learns and executes tasks through physical interaction between an intelligent agent and the environment. However, due to the diversity and complexity of dynamic changes of industrial scene tasks, the multi-task generalization ability of embodied intelligent robots in industrial scenes faces severe challenges. For example, in the scene of steel binding, it is difficult for an embodied intelligent robot to reasonably plan the multi-task execution sequence of "which to bind first, how to walk, and how to move the end". SUMMARY

[0003] Therefore, it is necessary to provide an embodied intelligent robot, a work sequence determination method and device, a computer readable storage medium and a computer program product, which are constrained by industry standards, target efficient work, and dynamically adapt, and can respond to the complex needs of the steel binding scene.

[0004] In a first aspect, the present application provides a work sequence determination method, comprising:

[0005] obtaining steel binding work data, and constructing a constraint parameter set according to the steel binding work data;

[0006] decomposing a global binding task into a plurality of independent intersection work units according to steel deployment positions, and assigning each intersection work unit with a corresponding work weight according to the constraint parameter set;

[0007] taking each intersection work unit carrying the work weight as a task node, and constructing a weighted semantic task graph between the task nodes according to the spatial position relationship between the intersections in the steel deployment positions;

[0008] dividing the weighted semantic task graph into a plurality of subgraphs by using a spatial clustering algorithm, planning and solving the work order in each subgraph based on a graph search method, and obtaining a subgraph work sequence corresponding to each subgraph;

[0009] connecting the subgraph work sequence corresponding to each subgraph based on the boundary task nodes between the subgraphs, and forming a global work sequence.

[0010] In one embodiment, the method comprises:

[0011] acquire the steel bar binding operation data from multidimensional knowledge of a preset database;

[0012] perform semantic segmentation and entity recognition processing on the steel bar binding operation data to obtain a binding object entity, an operation requirement entity, and a constraint condition entity in the steel bar binding operation data, and construct a triple relationship between the binding object entity, the operation requirement entity, and the constraint condition entity;

[0013] model the triple relationship to obtain a priority weight function and a timing logic formula;

[0014] form the constraint parameter set using the priority weight function and the timing logic formula.

[0015] In one embodiment, the spatial clustering algorithm is used to divide the weighted semantic task graph into a plurality of subgraphs, and the operation sequence in each subgraph is planned and solved based on a graph search method to obtain a subgraph operation sequence corresponding to each subgraph, including:

[0016] A spatial clustering algorithm is used to determine a region division strategy based on the intersection operation unit, and the weighted semantic task graph is divided into a plurality of subgraphs according to the region division strategy;

[0017] Each subgraph is respectively executed to perform the following operations for planning and solving:

[0018] The target task node in the subgraph whose operation weight satisfies a preset condition is found based on a graph search method, and a spatial adjacent task node of the target task node is determined;

[0019] An initial operation sequence corresponding to the subgraph is generated according to the target task node and the spatial adjacent task node;

[0020] The spatial coordinates of each task node corresponding to the subgraph and the initial operation sequence are used to calculate a total moving distance for performing the steel bar binding operation in the subgraph;

[0021] The initial operation sequence is iteratively optimized according to the total moving distance to obtain the subgraph operation sequence.

[0022] In one embodiment, the method further includes:

[0023] A spline curve of a mechanical arm motion trajectory is constructed according to the global operation sequence;

[0024] A preset target optimization function is used to solve the spline curve to obtain a motion planning trajectory, and the target optimization function is generated according to at least one of a geometric target, a dynamic target, and an efficiency target;

[0025] sensing real-time environment information in which the robot arm is located;

[0026] adjusting a target motion trajectory of the robot arm according to the real-time environment information and the motion planning trajectory, the target motion trajectory being a motion trajectory of the robot arm for performing the steel bar binding operation according to the global operation sequence.

[0027] In one of the embodiments, the adjusting the target motion trajectory of the robot arm according to the real-time environment information and the motion planning trajectory comprises:

[0028] determining a collision risk parameter of a current motion trajectory of the robot arm according to the real-time environment information and the motion planning trajectory;

[0029] in a case where the collision risk parameter is lower than a preset threshold, performing local trajectory correction on the current motion trajectory in a fine-tuning joint angle manner, and taking the corrected motion trajectory as the target motion trajectory;

[0030] in a case where the collision risk parameter is higher than the preset threshold, generating a local obstacle avoidance trajectory based on a current robot arm state and a target position, and connecting the local obstacle avoidance trajectory and the current motion trajectory as the target motion trajectory.

[0031] In one of the embodiments, the method further comprises:

[0032] generating a motion sequence of a robot arm end according to the global operation sequence;

[0033] obtaining motion parameters and execution constraint parameters of a basic motion unit of the robot arm end for performing the steel bar binding operation from a motion primitive library;

[0034] solving joint space motion parameters of the robot arm end by using the motion parameters and execution constraint parameters of the basic motion unit, combining a kinematics model of the robot arm and a preset physical constraint;

[0035] controlling the robot arm end to perform the steel bar binding operation according to the motion sequence according to the joint space motion parameters.

[0036] In a second aspect, the application further provides an operation sequence determination device, comprising:

[0037] a constraint construction module configured to obtain steel bar binding operation data, and construct a constraint parameter set according to the steel bar binding operation data;

[0038] a weight determination module, configured to decompose the global binding task into a plurality of independent intersection operation units according to the reinforcement bar deployment positions, and assign each of the intersection operation units with a corresponding operation weight according to the constraint parameter set;

[0039] a task graph construction module, configured to take each of the intersection operation units carrying the operation weight as a task node, and construct a weighted semantic task graph among the task nodes according to spatial positional relationships among intersections in the reinforcement bar deployment positions;

[0040] a sequence planning module, configured to divide the weighted semantic task graph into a plurality of subgraphs by using a spatial clustering algorithm, plan and solve an operation sequence in each of the subgraphs based on a graph search manner, and obtain a subgraph operation sequence corresponding to each of the subgraphs;

[0041] a sequence connection module, configured to connect the subgraph operation sequence corresponding to each of the subgraphs based on boundary task nodes among the subgraphs, and form a global operation sequence.

[0042] In a third aspect, the present application also provides a body-equipped intelligent robot. The body-equipped intelligent robot includes a memory and a processor, the memory stores a computer program, and the processor implements the operation sequence determination method in any one of the embodiments of the first aspect when executing the computer program.

[0043] In a fourth aspect, the present application also provides a computer readable storage medium. The computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the operation sequence determination method in any one of the embodiments of the first aspect.

[0044] In a fifth aspect, the present application also provides a computer program product. The computer program product includes a computer program, and the computer program is executed by a processor to implement the operation sequence determination method in any one of the embodiments of the first aspect.

[0045] The body-intelligent robot, the work sequence determination method, the device, the computer readable storage medium and the computer program product determine the work sequence of the steel bar binding work according to the steel bar binding work data, and the steel bar binding work data is used for constructing a constraint parameter set. The global binding task is decomposed into a plurality of independent intersection point work units according to the steel bar deployment position, and each intersection point work unit is given a corresponding work weight according to the constraint parameter set. Each intersection point work unit carrying the work weight is taken as a task node, and a weighted semantic task graph between the task nodes is constructed according to the spatial position relationship between the intersection points in the steel bar deployment position. The weighted semantic task graph is divided into a plurality of subgraphs by using a spatial clustering algorithm, the work order in each subgraph is planned and solved based on a graph search mode, and a subgraph work sequence corresponding to each subgraph is obtained. The subgraph work sequence corresponding to each subgraph is connected based on the boundary task nodes between the subgraphs, and a global work sequence is formed. The global work sequence not only realizes the fusion expression of'standard constraint-space information-task attribute' by using the weighted semantic task graph, and responds to the complex requirements of the steel bar binding scene with the industry standard as the constraint, the efficient work as the target and the dynamic adaptation as the core, but also adapts to the large-scale steel mesh scene, adopts the regional planning strategy, optimizes the internal sequence of the subgraph first, and then realizes the global coordination through the boundary node connection, which not only reduces the calculation complexity, but also ensures the continuity of the work in the region. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the present application or the related art. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, without creative labor, other related drawings can also be obtained from these drawings.

[0047] Figure 1 The flow environment diagram of the work sequence determination method in an embodiment;

[0048] Figure 2 The flowchart of the constraint parameter set construction step in an embodiment;

[0049] Figure 3 The flowchart of the subgraph work sequence determination step in an embodiment;

[0050] Figure 4 The flowchart of the target motion trajectory planning step in an embodiment;

[0051] Figure 5 The flowchart of the motion trajectory dynamic optimization step in an embodiment;

[0052] Figure 6A flowchart of steps of end-of-arm motion control in an embodiment;

[0053] Figure 7 A structural block diagram of the work sequence determination apparatus 700 in an embodiment. DETAILED DESCRIPTION

[0054] For the purpose, technical solutions and advantages of the present application to be clearer, the present application is further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0055] The data involved in the present application (including but not limited to data for analysis, stored data, displayed data, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0056] In an exemplary embodiment, as shown in Figure 1 A work sequence determination method in a steel bar binding operation scenario is provided, and the present embodiment takes the method applied to a somatic intelligent robot as an example. It should be understood that the method can also be applied to a control terminal, and can also be applied to a system including a terminal and a server, and is realized through the interaction of the terminal and the server. In the present embodiment, the method includes the following steps S102 to S110. Among them:

[0057] Step S102, obtaining steel bar binding operation data, and constructing a constraint parameter set according to the steel bar binding operation data.

[0058] In the present application, the steel bar binding operation in each embodiment refers to the operation of the somatic intelligent robot performing binding operation at the intersection of the steel bars to be bound by using its mechanical arm. The steel bar binding operation data can be used to represent multi-dimensional knowledge data such as construction specification, process guide and quality acceptance standard of the somatic intelligent robot when performing the steel bar binding operation.

[0059] The constraint parameter set can be used to represent the constraint conditions of the standardization and normalization of the steel bar binding operation. For example, the constraint parameter set can include explicit constraint parameters and implicit constraint parameters. The explicit constraint parameters are, for example, the binding priority of the force steel bar intersection point over the construction steel bar intersection point, the process sequence of the horizontal steel bar first and the vertical steel bar second, and the operation limit of the minimum binding spacing. The implicit constraint parameters involve potential requirement parameters such as construction efficiency of adjacent intersection point concentration operation and safety redundancy of avoiding mechanical arm motion dead zone.

[0060] For example, an embodied intelligent robot can obtain rebar tying operation data from a local database or data communication network. A graph-based knowledge representation method transforms the discrete rebar tying operation data into a computable set of constraint parameters.

[0061] Step S104: Based on the location of the rebar deployment, the global binding task is broken down into several independent intersection operation units, and each intersection operation unit is assigned a corresponding operation weight according to the constraint parameter set.

[0062] Among them, the location of the rebar deployment can be used to characterize geometric location information such as the spatial location and connectivity of the rebar intersections.

[0063] The global binding task refers to the binding task of all the intersections of the reinforcing bars that need to be bound in the current reinforcing bar binding scenario.

[0064] A cross-point work unit refers to the smallest work unit that takes the single cross-point of the reinforcing bars that need to be tied as the object of the reinforcing bar tying operation.

[0065] For example, an embodied intelligent robot can acquire the global rebar deployment location in a rebar tying scenario. Based on the rebar deployment location, it performs structured task allocation, breaking down the global tying task into several (may include one or more) sub-tasks targeting individual rebar intersections using a task decomposition algorithm. Based on these sub-tasks, it forms intersection work units corresponding to the individual rebar intersections that need to be tied. Based on information such as the spatial location, connectivity, spacing, rebar type (stressed rebar or structural rebar), and deployment direction (transverse or longitudinal rebar) of the rebar intersections corresponding to the intersection work units, it assigns a comprehensive priority, reachability, and operational complexity weight to each intersection work unit according to a set of constraint parameters.

[0066] Step S106: Treat each intersection work unit carrying the work weight as a task node, and construct a weighted semantic task graph between task nodes based on the spatial positional relationship between each intersection in the reinforcement deployment location.

[0067] For example, the embodied intelligent robot can treat each intersection work unit carrying a work weight as a task node. Based on the spatial relationships between the various rebar intersections in the rebar deployment location, such as spatial location, distance, and connectivity, and combined with the binding priority and process sequence in the work weights, a path is established between each task node. The task nodes are connected according to the established paths to construct a weighted semantic task graph between the task nodes.

[0068] Optionally, in some embodiments, the task nodes in the weighted semantic task graph can contain attribute information such as three-dimensional coordinates of the steel bar intersection points, binding types, priority weights, etc., the paths between the task nodes can represent the spatial adjacency relationship and process dependency relationship between the steel bar intersection points corresponding to the task nodes, and the weights can quantify the priority order of operation execution, realizing the fusion expression of “specification constraint—spatial information—task attribute”.

[0069] In step S108, the weighted semantic task graph is divided into a plurality of subgraphs by using a spatial clustering algorithm, and the operation sequence in each subgraph is planned and solved based on a graph search method to obtain a subgraph operation sequence corresponding to each subgraph.

[0070] For example, the embodied intelligent robot can calculate the distances between the task nodes in the weighted semantic task graph, cluster the task nodes with distances within a preset distance range by using a spatial clustering algorithm, obtain a cluster set of the task nodes, take an area where each cluster set is located as a subgraph, and thus divide the global weighted semantic task graph into a plurality of subgraphs. The operation sequence between the task nodes in each subgraph is planned and solved based on a graph search method such as a Dijkstra search method, an A* search method, etc., and an optimal solution that meets a preset planning target (such as the shortest moving distance, the highest work efficiency, etc.) is determined as a subgraph operation sequence corresponding to the current subgraph.

[0071] In step S110, the subgraph operation sequence corresponding to each subgraph is connected based on the boundary task nodes between the subgraphs to form a global operation sequence.

[0072] For example, the embodied intelligent robot can determine the spatial position area corresponding to the subgraph according to the spatial positions of the steel bar intersection points corresponding to each task node in the subgraph, and thus filter out the task nodes located at the area boundary as the boundary task nodes. The boundary task nodes between different subgraphs are matched, and the corresponding subgraphs are connected by using the boundary task nodes that are matched. Alternatively, the distances of the boundary task nodes between different subgraphs can be calculated, and the boundary task nodes with distances less than a preset distance threshold between different subgraphs are used to connect the corresponding subgraphs. Thus, all the subgraphs are connected, and the subgraph operation sequences of the subgraphs are connected based on the connection order to form a global operation sequence.

[0073] The above job sequence determination method obtains steel bar binding job data, constructs a constraint parameter set according to the steel bar binding job data, decomposes a global binding task into a plurality of independent intersection point job units according to the steel bar deployment position, and assigns a corresponding job weight to each intersection point job unit according to the constraint parameter set. Each intersection point job unit carrying a job weight is taken as a task node, and a weighted semantic task graph between the task nodes is constructed according to the spatial position relationship between the intersection points in the steel bar deployment position. The weighted semantic task graph is divided into a plurality of subgraphs by using a spatial clustering algorithm, the job sequence in each subgraph is planned and solved in a graph search manner, and a subgraph job sequence corresponding to each subgraph is obtained. The subgraph job sequence corresponding to each subgraph is connected based on the boundary task nodes between the subgraphs, and a global job sequence is formed. Not only can the weighted semantic task graph realize the fusion expression of “standard constraint—spatial information—task attribute”, but also can adapt to large-scale steel mesh scenes, adopt a regional planning strategy, optimize the internal sequence of the subgraph first, and then realize global coordination through the boundary nodes. This not only reduces the computational complexity, but also ensures the continuity of the job in the region.

[0074] In one exemplary embodiment, as shown in FIG. 2, step S102 can include steps S202 to S208. Among them: Figure 2

[0075] Step S202, obtaining steel bar binding job data from multi-dimensional knowledge in a preset database.

[0076] Among them, the preset database refers to a steel bar binding job industry standard database, which is used to store industry standard specification data.

[0077] Exemplarily, the embodied intelligent robot can obtain text information related to steel bar binding technical standard specifications from the multi-dimensional knowledge covered by the preset database, such as building construction specifications, steel bar binding process guidelines, and quality acceptance standards, as steel bar binding job data.

[0078] Step S204, performing semantic segmentation and entity recognition processing on the steel bar binding job data to obtain binding object entities, operation requirement entities, and constraint condition entities in the steel bar binding job data, and constructing a triple relationship between the binding object entities, the operation requirement entities, and the constraint condition entities.

[0079] ​Exemplarily, the embodied intelligent robot can perform semantic segmentation processing on the steel binding operation data to obtain a plurality of segmented text units. Entity recognition processing is performed on the text units obtained after the semantic segmentation processing to obtain binding object entities (such as force steel intersection points, structural steel intersection points, etc.), operation requirement entities (such as grasping first, binding later, and releasing last, etc.), and constraint condition entities (such as binding priority of force steel intersection points over structural steel intersection points, binding sequence of horizontal steel first and longitudinal steel later, operation limit of minimum binding spacing, and avoidance of mechanical arm motion dead zone, etc.) in the steel binding operation data. According to the triple structure of “binding object — operation requirement — constraint condition”, a triple relationship between the binding object entities, the operation requirement entities, and the constraint condition entities is constructed.

[0080] In step S206, a priority weight function and a timing logic formula are obtained according to the triple relationship modeling.

[0081] In step S208, a constraint parameter set is formed using the priority weight function and the timing logic formula.

[0082] Exemplarily, the embodied intelligent robot can perform constraint modeling according to the triple relationship, and convert the triple data into mathematical expressions such as the priority weight function and the timing logic formula (such as a timing logic formula for limiting the execution of the binding operation of the horizontal steel intersection point first and then the binding operation of the longitudinal steel intersection point). The mathematical expressions such as the priority weight function and the timing logic formula are used to form a constraint parameter set that can directly participate in the steel binding task allocation.

[0083] Optionally, in some embodiments, the following priority weight function formula can be modeled:

[0084]

[0085] wherein, represents the comprehensive priority weight of the i th steel intersection point. is a binding priority parameter. is an operation efficiency parameter. is a safety redundancy parameter. , , is a weight coefficient, and .

[0086] Optionally, in another embodiment, the embodied intelligent robot can also assign weights to the comprehensive priority, reachability, and operation complexity of each intersection point operation unit according to the above priority weight function formula.

[0087] ​In this embodiment, by building a complete technical link from specification analysis to task structured expression based on the industry standard database for rebar tying operations, the rebar tying task allocation driven by embodied intelligent robots is realized. This ensures that the subsequent rebar tying task allocation meets the technical requirements driven by embodied specifications from the source, thereby improving the standardization of the work sequence.

[0088] In one exemplary embodiment, such as Figure 3 As shown, step S108 may further include steps S302 to S310. Wherein:

[0089] Step S302: A spatial clustering algorithm is used to determine the region partitioning strategy based on the intersection point task unit. The weighted semantic task graph is divided into several subgraphs according to the region partitioning strategy.

[0090] For example, to adapt to large-scale steel mesh scenarios, the embodied intelligent robot can use a spatial clustering algorithm to cluster the spatial coordinates of each intersection work unit, treating the clustering results as a partitioned region to form a global region partitioning strategy. The weighted semantic task graph is then divided into several subgraphs corresponding one-to-one with the clustering results according to the region partitioning strategy.

[0091] Step S304: Perform the following operations for each subgraph to solve the planning problem: Find the target task node in the subgraph whose job weight meets the preset conditions based on graph search, and determine the spatially adjacent task nodes of the target task node.

[0092] For example, the embodied intelligent robot can perform the following planning and solving operations for each subgraph: find the task node in the current subgraph whose task weight meets the preset conditions (such as task node whose task weight is greater than the preset weight threshold or whose task weight is greater than 80%) based on graph search, and take the task node in the subgraph that is connected to the target task node as the spatial neighbor task node of the target task node.

[0093] Step S306: Generate the initial job sequence corresponding to the subgraph based on the target task node and spatially adjacent task nodes.

[0094] For example, an embodied intelligent robot can use a target task node as the first binding task in the initial work sequence. Based on the path, priority weight, and spatial distance between spatially neighboring task nodes and the target task node, the execution order of spatially neighboring task nodes after the target task node is determined. Then, using the spatially neighboring task node as a new target task node, new spatially neighboring task nodes are searched to determine the execution order corresponding to the new spatially neighboring task nodes, until all task nodes in the current subgraph have been traversed, thereby forming the initial work sequence corresponding to the current subgraph.

[0095] In step S308, the total moving distance of the embodied intelligent robot in the subgraph is calculated according to the spatial coordinates of each task node corresponding to the subgraph and the initial work sequence.

[0096] For example, the embodied intelligent robot can obtain the spatial coordinates of each task node in the subgraph. According to the execution order of each task node in the initial work sequence, the embodied intelligent robot calculates the sum of the moving distances from the spatial position of the first executed task node in the initial work sequence to the spatial position of the next executed task node, and so on, until the spatial position of the last executed task node, as the total moving distance of the embodied intelligent robot in the subgraph.

[0097] Optionally, in some embodiments, the total moving distance can be calculated by the following formula:

[0098]

[0099] wherein D represents the total moving distance, and n represents the total number of task nodes corresponding to the subgraph. represents the spatial position of the kth executed task node. represents the spatial position of the k+1th executed task node. represents the Euclidean distance between the kth executed task node and the k+1th executed task node.

[0100] In step S310, the initial work sequence is iteratively optimized according to the total moving distance to obtain a subgraph work sequence.

[0101] For example, the embodied intelligent robot can minimize the total moving distance as an optimization target to reduce the energy consumption of the embodied intelligent robot in the current subgraph for reinforcing bar binding work, iteratively optimize the initial work sequence based on a graph search method, and adjust the execution sequence order of each task node in the subgraph. Finally, when the number of iterations reaches a preset number threshold, or the total moving distance is lower than a preset distance threshold, the optimization is stopped, and the work sequence at this time is taken as the optimal subgraph work sequence of the current subgraph. Optionally, in some embodiments, during the iterative optimization of the work sequence, an optimization target of balancing the work load to avoid repeated work in local areas and an optimization target of completing the binding task of the frame reinforcing bar first and then processing the binding task of the detail node can also be added.

[0102] ​In this embodiment, by adopting a regional planning strategy, the global task graph is divided into several subgraphs using a spatial clustering algorithm, which can improve the sequence determination efficiency in a large-scale reinforcement mesh scene. By using a greedy strategy in each subgraph, the initial operation sequence is generated by preferentially selecting the target task node with high operation weight and its spatial adjacent nodes, and then the total moving distance is iteratively optimized and adjusted, which can reduce the energy consumption of the embodied intelligent robot in performing the reinforcement binding operation according to the subgraph operation sequence.

[0103] In one exemplary embodiment, as shown in Figure 4 The operation sequence determination method provided by the present application can further include the following steps S402 to S408. Among them:

[0104] Step S402, a spline curve of the mechanical arm motion trajectory is constructed according to the global operation sequence.

[0105] Exemplarily, the embodied intelligent robot can determine the order of the mechanical arm motion to each reinforcement intersection point for reinforcement binding operation according to the global operation sequence. According to the spatial position of each reinforcement intersection point and the binding operation sequence corresponding to the global operation sequence, the three-dimensional coordinates passed by the mechanical arm of the embodied intelligent robot from the first reinforcement intersection point to the last reinforcement intersection point in the process of performing reinforcement binding operation according to the global operation sequence are predicted, so as to form a spline curve of the mechanical arm.

[0106] Step S404, a preset target optimization function is used to solve the spline curve to obtain a motion planning trajectory.

[0107] Among them, the target optimization function is generated according to at least one of the geometric target, the dynamics target and the efficiency target. The geometric target usually requires the moving path of the mechanical arm to be smooth to ensure that the mechanical arm does not appear to move violently, and the posture to be continuous to ensure that the end effector of the mechanical arm is stable. The dynamics target usually needs to meet the joint speed and acceleration constraints of the mechanical arm to prevent overload. The efficiency target usually needs to minimize the motion time to improve the operation rhythm, shorten the end motion distance of the mechanical arm to improve the operation efficiency, etc. Alternatively, in some embodiments, the target optimization function can be a single-target optimization function generated by using any one of the geometric target, the dynamics target or the efficiency target. Or, in other embodiments, the target optimization function can also be obtained by fusing at least two of the geometric target, the dynamics target and the efficiency target through weighted summation.

[0108] Exemplarily, the embodied intelligent robot can pre-store a target optimization function. The preset target optimization function is used to optimize and solve the spline curve. For example, a polynomial curve can be constructed by using the target optimization function to dynamically constrain and locally adjust the spline curve to eliminate potential conflicts, so as to obtain a spline curve satisfying the dynamic constraint of the target optimization function. The trajectory representation of the spline curve is parameterized to form a motion planning trajectory of the robot arm.

[0109] Optionally, in some embodiments, in an application scenario where the operation time is strictly controlled, an efficiency target of the shortest operation time can be used to generate the currently used target optimization function. The embodied intelligent robot uses the target optimization function to solve the operation time corresponding to each point in the spline curve, thereby planning and generating a motion planning trajectory with the shortest operation time. Alternatively, in another embodiment, in an application scenario where the operation time of the robot arm end is concerned, an efficiency target of the shortest end trajectory path can also be used to construct the target optimization function. The embodied intelligent robot uses the target optimization function to operate and process the spline curve, so as to solve a motion planning trajectory with the shortest end trajectory path.

[0110] Optionally, in another embodiment, the trajectory planning task of the robot arm is heavy, and multiple targets such as the geometric target, the dynamics target and the efficiency target need to be balanced. A corresponding weight parameter can be set for each target. The multiple targets are converted into a single target form of the optimization function by using the weighted summation of the weight parameters, so as to obtain a target optimization function that can simultaneously consider multiple targets. For example, the weight parameter of the geometric target is set as a first parameter. The weight parameter of the dynamics target is set as a second parameter. The weight parameter of the efficiency target is set as a third parameter. Then the target optimization function = first parameter * geometric target + second parameter * dynamics target + third parameter * efficiency target. The values of the first parameter, the second parameter and the third parameter are 0-1, and the sum is 1. The higher the value of the weight parameter is, the more important the target corresponding to the weight parameter is in the decision of the robot arm motion trajectory planning.

[0111] Optionally, in some embodiments, the following cubic polynomial can be used to parameterize the robot arm joint space trajectory:

[0112] ,

[0113] wherein, is a function of the change of the robot arm joint angle with time t. 、 、 、 is a polynomial coefficient.

[0114] Step S406, sensing real-time environment information in which the robot arm is located.

[0115] Step S408: Adjust the target motion trajectory of the robotic arm according to real-time environmental information and motion planning trajectory.

[0116] Real-time environmental information may include information such as newly added obstacles and deviations in the position of reinforcing bars.

[0117] The target motion trajectory is the motion trajectory of the robotic arm performing the rebar binding operation according to the global operation sequence.

[0118] For example, during the actual rebar tying operation performed by the embodied intelligent robot, real-time environmental information about the robot's robotic arm is perceived. Based on this real-time environmental information, it is determined whether a collision hazard exists while moving along the planned motion trajectory. If the probability of a collision is high, the position of the robotic arm within the planned motion trajectory is adjusted, thereby changing the arm's trajectory to avoid collisions with obstacles in the current environment. The adjusted trajectory is then used as the target trajectory for the robotic arm to perform the rebar tying operation according to the global work sequence.

[0119] In this embodiment, spline curves of the embodied intelligent robot arm are generated based on the global operation sequence. During the trajectory planning process of the robot arm, the motion planning trajectory of the robot arm is generated by solving the objective optimization function generated based on at least one of the geometric objective, dynamic objective, and efficiency objective. Then, the motion trajectory planning is dynamically adjusted by combining the perceived real-time environmental information. This not only retains the high efficiency of the global motion trajectory planning, but also adjusts the actual target motion trajectory through dynamic response to environmental perception, thereby improving the adaptability to environmental changes.

[0120] In one exemplary embodiment, such as Figure 5 As shown, step S408 may further include steps S502 to S506. Wherein:

[0121] Step S502: Determine the collision risk parameters of the current motion trajectory of the robotic arm based on real-time environmental information and motion planning trajectory.

[0122] For example, an embodied intelligent robot can use a geometric constraint detection algorithm to combine the planned trajectory position in the motion planning trajectory with information such as the position of obstacles and the deviation of steel bars in the real-time environment to determine whether there is a collision risk in the current motion trajectory, thereby obtaining the collision risk parameters of the current motion trajectory of the robotic arm.

[0123] Step S504: If the collision risk parameter is lower than the preset threshold, the current motion trajectory is locally corrected by fine-tuning the joint angle, and the corrected motion trajectory is used as the target motion trajectory.

[0124] Exemplarily, the embodied intelligent robot can extract a local trajectory that is likely to collide with the obstacle from the current motion trajectory when the collision risk parameter is lower than the preset threshold, and take the joint space trajectory of the mechanical arm corresponding to the local trajectory as a reference curve. A small displacement is applied to the end effector of the mechanical arm based on the preset fine-tuning logic, and the small displacement is converted into an incremental joint angle. The trajectory points of the reference curve are corrected based on the target optimization function and the incremental joint angle, thereby obtaining a corrected motion trajectory of the mechanical arm, and the corrected motion trajectory is taken as a target motion trajectory to guide the subsequent motion of the mechanical arm.

[0125] In step S506, when the collision risk parameter is higher than the preset threshold, a local obstacle avoidance trajectory is generated based on the current state of the mechanical arm and the target position, and the local obstacle avoidance trajectory is connected with the current motion trajectory as a target motion trajectory.

[0126] The state of the mechanical arm can be used to describe the spatial position, motion trend and other state information of the mechanical arm.

[0127] The target position can be used to represent the spatial position of the next steel bar intersection point that needs to perform the steel bar binding operation in the motion planning trajectory.

[0128] Exemplarily, when the collision risk parameter is higher than the preset threshold, the dynamic path re-planning logic of the mechanical arm is triggered. The embodied intelligent robot can generate a local obstacle avoidance trajectory that moves from the current position of the mechanical arm to the target position and bypasses the obstacle position in the real-time environmental information based on the current state of the mechanical arm and the target position. The local obstacle avoidance trajectory is seamlessly connected with the current motion trajectory of the mechanical arm by using trajectory smoothing transition technology, and the connected motion trajectory is taken as a target motion trajectory.

[0129] In this embodiment, whether the pre-planned motion trajectory of the mechanical arm has a collision risk is judged based on real-time environmental information. In the case of low risk, the target motion trajectory is obtained by locally correcting the joint angle. In the case of high risk, a local obstacle avoidance trajectory is generated based on the current state of the mechanical arm and the target position, and the new local obstacle avoidance trajectory is seamlessly connected with the old current motion trajectory to obtain a target motion trajectory. The motion planning trajectory can be dynamically and real-timely adjusted in a complex environment in combination with real-time environmental information, and the real-time performance and stability of the local safety obstacle avoidance and dynamic path adjustment algorithm are ensured.

[0130] In one exemplary embodiment, as shown in Figure 6 The work sequence determination method provided by the present application can further include the following steps S602 to S608. Wherein:

[0131] In step S602, an action sequence of the end of the mechanical arm is generated according to the global work sequence.

[0132] Exemplarily, the embodied intelligent robot can construct a behavior tree according to the global operation sequence, take the behavior tree as a logic control core for the end of the robot arm to perform the steel binding task, and based on the global operation sequence planned by the upper layer and the requirements of the steel binding task, logically arrange the actions of the end of the robot arm, and generate the action sequence of the end of the robot arm. Optionally, in other embodiments, the embodied intelligent robot can also realize state monitoring of the end of the robot arm according to the behavior tree, timely discover abnormal conditions of the end of the robot arm in performing the steel binding operation and perform fault recovery based on the behavior tree, and ensure that the task is performed according to the predetermined global operation sequence.

[0133] In step S604, the motion parameters and execution constraint parameters of the basic action units of the end of the robot arm for performing the steel binding operation are obtained from the action primitive library.

[0134] In step S606, the motion parameters and execution constraint parameters of the basic action units are used in combination with the kinematics model of the robot arm and the preset physical constraints to solve the joint space motion parameters of the end of the robot arm.

[0135] In step S608, the end of the robot arm is controlled to perform the steel binding operation according to the action sequence according to the joint space motion parameters.

[0136] The action primitive library can include parameterizable basic action units based on the disassembly of the binding process, such as grasping, binding, releasing, etc. Each basic motion unit in the action primitive library includes motion parameters and execution constraint parameters. The motion parameters can be used to represent the pose, angle, motion direction, etc. when performing the action corresponding to the basic motion unit. The execution constraint parameters can be used to represent the constraint conditions when performing the action corresponding to the basic motion unit, such as joint angle activity range, obstacle avoidance distance, motion accuracy, etc.

[0137] Exemplarily, the embodied intelligent robot can obtain the motion parameters and execution constraint parameters of each basic action unit in the process of the end of the robot arm performing the steel binding operation from the preset action primitive library. The motion parameters and execution constraint parameters of the basic action units are adjusted and combined according to the specific operation requirements under the actual working conditions. The constraint type redundant inverse kinematics is used as the bottom layer motion control technology, the motion parameters and execution constraint parameters of each basic action unit are used, the kinematics model of the robot arm and the preset physical constraints of the end of the robot arm are used to solve the joint space motion parameters that meet the preset end of the robot arm pose requirements. The joint space motion parameters are mapped to dynamic action parameters to control the end of the robot arm to perform the corresponding steel binding operation according to the globally planned action sequence.

[0138] Optionally, in some embodiments, the relationship between the end-of-arm pose and the joint angle can be mathematically modeled with reference to the following formula:

[0139] ,

[0140] where T is the pose matrix of the end-of-arm effector. denotes the joint angle of each joint of the robot arm. denotes the forward kinematics function of the robot arm.

[0141] In this embodiment, by constructing a three-level collaborative mechanism of "behavior tree-action primitive library-constraint type redundant inverse kinematics", precise control and flexible adjustment of the end-of-arm action are realized. The behavior tree is used as the logical control core of the robot arm end executing the steel binding task. Based on the global operation sequence planned by the upper layer and the requirements of the steel binding task, the logical arrangement of the end-of-arm action sequence is realized, ensuring that the robot arm end executes the steel binding operation according to the preset process. According to the actual working condition, the motion parameters and execution constraint parameters of each basic action unit in the action primitive library are adjusted and combined, and the robot arm end is controlled according to the joint space motion parameters obtained by solving. The action can ensure the accuracy of the action while avoiding the collision of the robot arm joints and itself, realizing the efficient and robust execution of the end-of-arm action. At the same time, through the cooperation of the three-level collaborative mechanism, the complex binding process can be converted into an ordered and optimized parameterized action sequence, ensuring the precision and adaptability of the end-of-arm action.

[0142] Optionally, in some embodiments, the embodied intelligent robot can run a work sequence intelligent arrangement trajectory optimization system. Through this system, the work sequence determination method provided in each of the above embodiments is executed. The industry standard for steel binding operation is used as a constraint, efficient operation is used as a goal, dynamic adaptation is used as a core, and the order optimization multi-objective trajectory planning and dynamic obstacle avoidance are guided through the specification analysis and semantic weighted task graph construction, further realizing the three-level collaborative mechanism of the robot arm end, and realizing the full-process intelligentization of the steel binding task allocation and robot arm trajectory optimization under the embodied specification driven technical system.

[0143] It should be understood that although the steps in the flowcharts involved in the above embodiments are shown in sequence according to the arrows, the steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, the execution of the steps is not strictly limited in sequence, and the steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the above embodiments can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of the steps or stages is not necessarily sequential, but can be alternately executed with at least some of the other steps or steps or stages in other steps. It can be understood that the steps in different embodiments can be freely combined as needed, and various non-contradictory schemes formed by the combination are within the scope of protection of the present application.

[0144] Based on the same inventive concept, the embodiments of the present application also provide a job sequence determination device for implementing the above-mentioned job sequence determination method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more job sequence determination device embodiments provided below can refer to the limitations of the job sequence determination method in the above text, which will not be repeated here.

[0145] In one exemplary embodiment, as shown in Figure 7 A job sequence determination device 700 is provided, comprising a constraint construction module 702, a weight determination module 704, a task graph construction module 706, a sequence planning module 708, and a sequence connection module 710, wherein:

[0146] The constraint construction module 702 is configured to obtain reinforcement binding job data and construct a constraint parameter set according to the reinforcement binding job data.

[0147] The weight determination module 704 is configured to decompose a global binding task into a plurality of independent intersection point job units according to the reinforcement deployment position, and assign a corresponding job weight to each intersection point job unit according to the constraint parameter set.

[0148] The task graph construction module 706 is configured to take each intersection point job unit carrying a job weight as a task node, and construct a weighted semantic task graph between the task nodes according to the spatial position relationship between the intersections in the reinforcement deployment position.

[0149] The sequence planning module 708 is configured to divide the weighted semantic task graph into a plurality of subgraphs using a spatial clustering algorithm, plan and solve the job order in each subgraph based on a graph search method, and obtain a subgraph job sequence corresponding to each subgraph.

[0150] The sequence connection module 710 is used to connect the subgraph job sequences corresponding to each subgraph based on the boundary task nodes between subgraphs to form a global job sequence.

[0151] In an exemplary embodiment, the constraint construction module 702 is further configured to: obtain rebar tying operation data from multi-dimensional knowledge in a preset database; perform semantic segmentation and entity recognition processing on the rebar tying operation data to obtain tying object entities, operation requirement entities, and constraint condition entities in the rebar tying operation data; construct triplet relationships between tying object entities, operation requirement entities, and constraint condition entities; model the priority weight function and temporal logic formula based on the triplet relationships; and form a constraint parameter set using the priority weight function and temporal logic formula.

[0152] In an exemplary embodiment, the sequence planning module 708 is further configured to: determine a region partitioning strategy based on intersection work units using a spatial clustering algorithm; divide the weighted semantic task graph into several subgraphs according to the region partitioning strategy; and perform the following planning and solving operations on each subgraph: find target task nodes in the subgraph whose work weights meet preset conditions using graph search; determine spatially adjacent task nodes of the target task nodes; generate an initial work sequence corresponding to the subgraph based on the target task nodes and spatially adjacent task nodes; calculate the total moving distance for performing rebar tying work within the subgraph based on the spatial coordinates of each task node corresponding to the subgraph and the initial work sequence; and iteratively optimize the initial work sequence based on the total moving distance to obtain the subgraph work sequence.

[0153] In an exemplary embodiment, the job sequence determination device 700 further includes a trajectory planning module, which is used to construct a spline curve of the robotic arm's motion trajectory based on the global job sequence; solve the spline curve using a preset target optimization function to obtain the motion planning trajectory, wherein the target optimization function is generated based on at least one of geometric, dynamic, and efficiency targets; sense the real-time environmental information of the robotic arm; and adjust the target motion trajectory of the robotic arm based on the real-time environmental information and the motion planning trajectory, wherein the target motion trajectory is the motion trajectory of the robotic arm performing rebar binding work according to the global job sequence.

[0154] In an exemplary embodiment, the trajectory planning module is further configured to: determine the collision risk parameters of the current motion trajectory of the robotic arm based on real-time environmental information and the motion planning trajectory; if the collision risk parameters are lower than a preset threshold, perform local trajectory correction on the current motion trajectory by fine-tuning the joint angles, and use the corrected motion trajectory as the target motion trajectory; if the collision risk parameters are higher than the preset threshold, generate a local obstacle avoidance trajectory based on the current state of the robotic arm and the target position, and connect the local obstacle avoidance trajectory with the current motion trajectory as the target motion trajectory.

[0155] In an example embodiment, the job sequence determination apparatus 700 further comprises a motion control module configured to generate a motion sequence of the robot end effector according to the global job sequence; obtain motion parameters of a basic motion unit of the robot end effector performing the reinforcement binding job and execution constraint parameters from the action primitive library; solve joint space motion parameters of the robot end effector by using the motion parameters of the basic motion unit and the execution constraint parameters, combining a kinematics model of the robot and preset physical constraints; and control the robot end effector to perform the reinforcement binding job according to the motion sequence based on the joint space motion parameters.

[0156] The modules in the job sequence determination apparatus 700 described above can be implemented by software, hardware, or a combination thereof, in whole or in part. The modules described above can be embedded in or independent of the processor in the embodied intelligent robot in hardware form, or stored in the memory in the embodied intelligent robot in software form, so as to be called and executed by the processor to perform the operations corresponding to the modules.

[0157] In an example embodiment, an embodied intelligent robot is provided, comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the method embodiments described above when executing the computer program.

[0158] In an example embodiment, a computer readable storage medium is provided, storing a computer program, and the computer program is executed by a processor to implement the steps in the method embodiments described above.

[0159] In an example embodiment, a computer program product is provided, comprising a computer program, and the computer program is executed by a processor to implement the steps in the method embodiments described above.

[0160] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., without being limited thereto.

[0161] The technical features of the above embodiments can be combined arbitrarily. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present application.

[0162] The above-described embodiments are merely illustrative of several embodiments of the present application, which are described in more detail and in a specific manner, but should not be construed as limiting the scope of the patent of the present application. It should be noted that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A work sequence determination method characterized by comprising: The method comprises: obtaining steel binding operation data, and constructing a constraint parameter set according to the steel binding operation data; dissolving a global binding task into a plurality of independent intersection operation units according to steel deployment positions, and assigning a corresponding operation weight to each intersection operation unit according to the constraint parameter set; taking each intersection operation unit carrying the operation weight as a task node, and constructing a weighted semantic task graph between the task nodes according to the spatial position relationship between intersections in the steel deployment positions; dividing the weighted semantic task graph into a plurality of subgraphs by using a spatial clustering algorithm, planning and solving the operation sequence in each subgraph by using a graph search method, and obtaining a subgraph operation sequence corresponding to each subgraph; connecting the subgraph operation sequence corresponding to each subgraph based on the boundary task nodes between the subgraphs, and forming a global operation sequence; wherein the obtaining of the steel binding operation data and the construction of the constraint parameter set according to the steel binding operation data comprise: obtaining the steel binding operation data from multidimensional knowledge in a preset database; performing semantic segmentation and entity recognition processing on the steel binding operation data to obtain binding object entities, operation requirement entities and constraint condition entities in the steel binding operation data, and constructing a triple relationship between the binding object entities, the operation requirement entities and the constraint condition entities; modeling the triple relationship to obtain a priority weight function and a timing logic formula; forming the constraint parameter set by using the priority weight function and the timing logic formula; wherein the dividing of the weighted semantic task graph into a plurality of subgraphs by using a spatial clustering algorithm, the planning and solving of the operation sequence in each subgraph by using a graph search method, and the obtaining of a subgraph operation sequence corresponding to each subgraph comprise: determining a region division strategy based on the intersection operation units by using a spatial clustering algorithm, and dividing the weighted semantic task graph into a plurality of subgraphs according to the region division strategy; respectively performing the following operations on each subgraph for planning and solving: finding a target task node in the subgraph whose operation weight meets a preset condition by using a graph search method, and determining spatial adjacent task nodes of the target task node; generating an initial operation sequence corresponding to the subgraph according to the target task node and the spatial adjacent task nodes; calculating a total moving distance for performing the steel binding operation in the subgraph according to the spatial coordinates of each task node of the subgraph and the initial operation sequence; iteratively optimizing the initial operation sequence according to the total moving distance to obtain the subgraph operation sequence.

2. The method of claim 1, wherein, The method further comprises: constructing a spline curve of a mechanical arm motion trajectory according to the global operation sequence; solving the spline curve by using a preset target optimization function to obtain a motion planning trajectory, the target optimization function being generated according to at least one of a geometric target, a dynamics target and an efficiency target; sensing real-time environmental information in which the mechanical arm is located; Adjust a target motion trajectory of the mechanical arm according to the real-time environment information and the motion planning trajectory, the target motion trajectory being a motion trajectory of the mechanical arm for performing the steel bar binding operation according to the global operation sequence.

3. The method of claim 2, wherein, The adjusting of the target motion trajectory of the mechanical arm according to the real-time environment information and the motion planning trajectory comprises: determining a collision risk parameter of a current motion trajectory of the mechanical arm according to the real-time environment information and the motion planning trajectory; in a case where the collision risk parameter is lower than a preset threshold, performing local trajectory correction on the current motion trajectory in a manner of fine-tuning joint angles, and taking the corrected motion trajectory as the target motion trajectory; in a case where the collision risk parameter is higher than the preset threshold, generating a local obstacle avoidance trajectory based on a current mechanical arm state and a target position, and connecting the local obstacle avoidance trajectory and the current motion trajectory as the target motion trajectory.

4. The method of claim 2, wherein, The method further comprises: generating a motion sequence of an end of the mechanical arm according to the global operation sequence; obtaining motion parameters and execution constraint parameters of a basic motion unit of the end of the mechanical arm for performing the steel bar binding operation from a motion primitive library; solving joint space motion parameters of the end of the mechanical arm by using the motion parameters and the execution constraint parameters of the basic motion unit, combining a kinematics model of the mechanical arm and a preset physical constraint; and controlling the end of the mechanical arm to perform the steel bar binding operation according to the motion sequence according to the joint space motion parameters.

5. A work sequence determination device characterized by comprising: The device comprises: a constraint construction module configured to obtain steel bar binding operation data and construct a constraint parameter set according to the steel bar binding operation data; a weight determination module configured to decompose a global binding task into a plurality of independent cross-point operation units according to steel bar deployment positions, and assign a corresponding operation weight to each cross-point operation unit according to the constraint parameter set; a task graph construction module configured to take each cross-point operation unit carrying the operation weight as a task node, and construct a weighted semantic task graph between the task nodes according to spatial position relationships between cross points in the steel bar deployment positions; a sequence planning module configured to divide the weighted semantic task graph into a plurality of subgraphs by using a spatial clustering algorithm, plan and solve an operation sequence in each subgraph by using a graph search method, and obtain a subgraph operation sequence corresponding to each subgraph; a sequence connection module configured to connect the subgraph operation sequence corresponding to each subgraph based on boundary task nodes between the subgraphs, and form a global operation sequence. The constraint construction module is further configured to acquire the steel bar binding operation data from multidimensional knowledge of a preset database, perform semantic segmentation and entity recognition processing on the steel bar binding operation data, obtain binding object entities, operation requirement entities, and constraint condition entities in the steel bar binding operation data, construct a triple relationship between the binding object entities, the operation requirement entities, and the constraint condition entities, model the triple relationship to obtain a priority weight function and a time sequence logic formula, and form the constraint parameter set by using the priority weight function and the time sequence logic formula. The sequence planning module is further configured to determine a region division strategy based on the intersection operation unit by using a spatial clustering algorithm, divide the weighted semantic task graph into a plurality of subgraphs according to the region division strategy, and perform the following operations on each subgraph to plan and solve: finding a target task node in the subgraph based on a graph search manner, the task weight of the target task node satisfying a preset condition, determining a spatial adjacent task node of the target task node, generating an initial operation sequence corresponding to the subgraph according to the target task node and the spatial adjacent task node, calculating a total moving distance of the steel bar binding operation in the subgraph according to the spatial coordinates of each task node of the subgraph and the initial operation sequence, and iteratively optimizing the initial operation sequence according to the total moving distance to obtain a subgraph operation sequence.

6. The apparatus of claim 5, wherein, The device further includes a trajectory planning module configured to construct a spline curve of a mechanical arm motion trajectory according to the global operation sequence, solve the spline curve by using a preset target optimization function, obtain a motion planning trajectory, the target optimization function being generated according to at least one of a geometric target, a dynamics target, and an efficiency target, perceive real-time environment information of the mechanical arm, and adjust a target motion trajectory of the mechanical arm according to the real-time environment information and the motion planning trajectory, the target motion trajectory being a motion trajectory of the mechanical arm performing the steel bar binding operation according to the global operation sequence.

7. The apparatus of claim 6, wherein, The trajectory planning module is further configured to determine a collision risk parameter of a current motion trajectory of the mechanical arm according to the real-time environment information and the motion planning trajectory, perform local trajectory correction on the current motion trajectory by adjusting a joint angle in a fine-tuning manner in a case where the collision risk parameter is lower than a preset threshold, take the corrected motion trajectory as the target motion trajectory, and generate a local obstacle avoidance trajectory based on a current mechanical arm state and a target position in a case where the collision risk parameter is higher than the preset threshold, and splice the local obstacle avoidance trajectory and the current motion trajectory as the target motion trajectory.

8. A somatic intelligent robot comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 4.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 4.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 4. The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 4.

Images